Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L1/00—Arrangements for detecting or preventing errors in the information received
  • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
  • H04L1/0056—Systems characterized by the type of code used
  • H04L1/0059—Convolutional codes
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L1/00—Arrangements for detecting or preventing errors in the information received
  • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
  • H04L1/0056—Systems characterized by the type of code used
  • H04L1/0064—Concatenated codes
  • H04L1/0066—Parallel concatenated codes

Definitions

• the invention relates to communication systems. More particularly, the invention relates to methods and apparatus for improving reception and decoding of signals encoded using turbo codes.
• CDMA code division multiple access
• CDMA modulation techniques can provide capacity improvements over other techniques, such as time division multiple access (TDMA) and frequency division multiple access (FDMA) based in part on CDMA's use of orthogonal functions or codes. Additionally, CDMA receivers employ Viterbi decoders that employ Viterbi algorithms to perform maximum likelihood decoding of received signals.
• TDMA time division multiple access
• FDMA frequency division multiple access
• CDMA receivers employ Viterbi decoders that employ Viterbi algorithms to perform maximum likelihood decoding of received signals.
• a new encoding scheme employing "turbo" codes employs a combination of two simple encoders that receive a block of K information bits, and which generate parity symbols from two simple recursive convolutional codes, each having a small number of states.
• the K information bits are sent uncoded, with the parity symbols, over a noisy channel.
• an interleaver permutes the original K information bits before inputting such bits to the second encoder.
• the permutation causes one encoder to produce low-weight code words, while the other encoder produces high- weight code words.
• the resulting code is similar to "random" block codes with K information bits. Random block codes are known to achieve Shannon-limit performance when K is large, but truly random block codes require a prohibitively costly and complex decoding algorithm.
• a pair of simple decoders employing iterative maximum a posteriori algorithms receive both the original K information bits, together with one of the sets of parity symbols from one of the two encoders.
• the decoders are individually matched to the simple codes, and each decoder sends an a posteriori likelihood estimate of the decoded bits to the other decoder, while using the corresponding estimates from the other decoder as an a priori likelihood.
• Uncommon information bits corrupted by the noisy channel are available to each decoder to minimize the a priori likelihoods, while the decoders use maximum a priori decoding algorithms, which requires the same number of states as the Viterbi algorithm.
• turbo codes may be found, for example, in "A Primer On Turbo Code Concepts," by B. Sklar, IEEE Communications Magazine, 35:12 (December 1997).
• turbo codes have the potential to improve information transmission over a noisy channel by a large amount, compared to classical Viterbi decoding, even results remarkably close to the theoretical Shannon-limit.
• idealized conditions at the receiver which are in practice, very difficult to approach.
• One troublesome idealized condition is that the receiver is presumed to know the signal and noise power on a per symbol basis. This can be particularly difficult because the signals are multiplied by information bits, which makes the means or averages of the signals zero, thus disallowing usual averaging procedures for estimating signal noise power.
• turbo encoding is much less attractive, if not worse, when compared to conventional Viterbi decoding of a convolutional code.
• the inventors have found, through experimentation, that classical techniques for estimating channel conditions induce a loss in performance because of the poor performance of such estimation techniques.
• the inventors have found a new set of techniques for signal and noise power estimation that apply not only to turbo codes, but to applications in other fields where such type of estimation is needed (e.g., power control in CDMA telecommunications systems, other concatenated coding schemes, etc.).
• aspects of the invention include a method for estimating channel conditions of received signals.
• the method includes: (a) receiving a signal encoded with concatenated codes over a channel having noise, wherein the received signal has a certain amplitude; (b) estimating a certain amplitude based on the received signal; and (c) separately estimating a variance ⁇ of the noise, based on the received pilot signal.
• FIG. 1 is a simplified block diagram of a wireless communications system employing the invention.
• FIG. 2 is a simplified block diagram of a transceiver in the wireless communication system of FIG. 1, in accordance with embodiments of the invention.
• FIG. 3 is a simplified block diagram of a receiver in the transceiver of FIG. 2, in accordance with embodiments of the invention.
• FIG. 4 A is a plot of energy verses symbol number for a CDMA mobile demodulator in direct spread mode, moving at 3 km/h in a classical Rayleigh fading environment, using a classical method of estimating energy per symbol.
• FIG. 4B is a plot of energy verses symbol number for a CDMA mobile demodulator in direct spread mode, moving at 3 km/h in a classical Rayleigh fading environment, using a classical method of estimating energy per symbol.
• FIG. 4B is a plot of energy verses symbol number for a
• CDMA mobile demodulator in direct spread mode, moving at 3 km/h in a classical Rayleigh fading environment, using one embodiment of the invention.
• FIG. 4C is a plot of a reference energy curve showing energy per symbol in the fading environment of FIGs. 4A and 4B.
• FIG. 5 is a plot of a function g(E(x 2 )/E( ⁇ x ⁇ ).
• FIG. 6 is a simplified flow chart of a process for estimating channel performance performed by the receiver of FIG. 3.
• FIG. 1 illustrates an exemplary cellular subscriber communication system 100, which uses multiple access techniques such as CDMA for communicating between users of user stations (e.g., mobile telephones) and cell-sites or base stations.
• a mobile user station 102 communicates with a base station controller 104 through one or more base stations 106a, 106b, etc.
• a fixed user station 108 communicates with the base station controller 104, but through only one or more predetermined and proximate base stations, such as the base stations 106a and 106b.
• the base station controller 104 is coupled to and typically includes interface and processing circuitry for providing system control to the base stations 106a and 106b.
• the base station controller 104 may also be coupled to and communicate with other base stations 106a and 106b, and possibly even other base station controllers.
• the base station controller 104 is coupled to a mobile switching center 110, which in turn is coupled to a home location register 112.
• the base station controller 104 and the mobile switching center 110 compare registration signals received from the user stations 102 or 108 to data contained in the home location register 112, as is known in the art. Soft handoffs may occur between the base station controller 104 and other base station controllers, and even between the mobile switching center 110 and other mobile switching centers, as is known by those skilled in the art.
• the base station controller 104 When the system 100 processes telephone or data traffic calls, the base station controller 104 establishes, maintains and terminates the wireless link with the mobile station 102 and the fixed user station 108, while the mobile switching center 110 establishes, maintains and terminates communications with a public switched telephone network (PSTN). While the discussion below focuses on signals transmitted between the base station 106a and the mobile station 102, those skilled in the art will recognize that the discussion equally applies to other base stations, and to the fixed user station 108.
• PSTN public switched telephone network
• FIG. 2 is a simplified block diagram of a portion of a transceiver 200 for use in either the base station 106a or 106b or the user stations 102 or 108 in the wireless communication system 100 of FIG. 1, under embodiments of the invention.
• the transceiver 200 includes a transmitter 202 and receiver 204 sharing an antenna 210 that transmits and receives signals to and from other transceivers 200.
• a duplexer 212 separates received signals from signals being transmitted by the transmitter 202 and routes the received signals to the receiver 204.
• the receiver 204 frequency shifts, demodulates and decodes the received signal. For example, the receiver 204 converts received signals to either baseband or an intermediate frequency and performs Walsh code demodulation, and also performs power and signal quality measurements.
• a control processor 216 provides much of the processing of the received signal, as described below.
• a memory 218 permanently stores routines performed by the control processor 216, and provides a temporary storage of data such as received frames.
• the transmitter 202 encodes, modulates, amplifies and up converts signals to be transmitted.
• the transmitter 202 forms a forward traffic link data signal for re-transmission of received signals by the base stations 106a or 106b to the user stations 102 or 108. In another embodiment, the transmitter 202 forms a reverse link traffic data signal for transmission from the user stations 102 or 108 back to the base station 106a.
• the receiver system 204 provides decoded received data to the user, and accepts information for transmission through the transmitter system 202 from the user, via an input/ output (I/O) module 222 coupled to the control processor 216.
• I/O input/ output
• a turbo decoder 300 which forms part of the receiver system 204, is shown.
• the decoder 300 may form part of the control processor 216 or the control processor may perform the operations described below for the decoder 300.
• the decoder 300 includes a signal and noise estimator 302 that receives an input channel having information and two parity signals or channels.
• the signal and noise estimator 302 separately estimates the power of the signal and noise in the received input channel, as described more fully below.
• a pair of simple decoders 304 and 306 each receive the information signal from the input channel. Additionally, the first decoder 304 (Decoder 1) receives the first parity symbol, while the second decoder 306 (Decoder 2) receives the second parity symbol.
• the decoder 300 (and the transceiver 200) may exist in the mobile station 102, and receive the information and parity symbols from the base station 106a.
• a typical turbo encoder at the base station 106a (not shown) provides a pair of simple encoders that generate parity signals from two simple recursive convolutional codes, where such codes have a small number of states.
• the parity symbols produced by the first encoder are received by the first decoder 304, while the parity symbols produced by the second encoder are received by the second decoder 306.
• the decoders 304 and 306 are matched to the codes of their respective encoders.
• the first decoder 304 provides an a posteriori likelihood estimate of decoded bits on the information channel to the second decoder 306 over line 308.
• the second decoder 306 does likewise over the line 310 for its corresponding estimate.
• the a posteriori estimates are used as a priori likelihoods for each decoder. Several iterations are performed until reaching a satisfactory convergence, at which point the final output of the likelihood estimates are provided.
• the signal and noise estimator 302 separately analyzes signal and noise on the received input channel to generate appropriate estimates.
• the input signal x to the decoder 300 may be represented as:
• Equation (1) may be rewritten by replacing the amplitude -4 with the square root of the energy per symbol (i.e., JE ⁇ ).
• the estimator 302 must provide separate estimates for the signal amplitude or energy «jE si and the noise variance ⁇ 2 . If the signal b t were always equal to one, the estimator 302 could readily measure the sample mean and the sample variance. However, since ⁇ b ⁇ is equal to ⁇ 1, then the mean is effectively zero.
• the estimator 302 may use the sample mean of the magnitude of x's, under the following equations:
• E ⁇ can be estimated as: XX (4)
• a second negative order polynomial curve fit may be used for / as a function of E s / N t .
• E(X 2 )-E( ⁇ X ⁇ ) 2 I l— ⁇ 2 and — > 1, so the SNR in dB is greater than x ' ⁇ j 1-2/ ⁇ ° zero and is not equal to minus infinity.
• the system controls power based on estimating current received energy per symbol values. Power control attempts keep the received signal to noise ratio constant, and therefore the ratio — — — varies little. Furthermore, the lower half of equation (5)
• FIG. 4C shows a reference curve of amplitude verses quadrature phase shift key (QPSK) symbols 1-5,000 in a Rayleigh fading environment with the decoder moving at a rate of 3 km/h.
• the demodulator in this example is demodulating under CDMA techniques ("CDMA 2000") in Direct Spread mode.
• FIG. 4A shows a typical method of estimating amplitude and noise using classical estimations of the mean of the received signal.
• FIG. 4B shows amplitude verses QPSK symbol under the above method, where noise ⁇ 2 and signal amplitude A are estimated separately.
• the curve of FIG. 4B closely follows the reference curve of FIG. 4C, while the curve of FIG. 4 A fails to track the reference curve well.
• a deep fade between about symbols 4500 and 4700 in the reference curve of FIG. 4C is tracked quite well under the above method as shown in FIG. 4B.
• such a deep fade results in an opposite result under the classical method shown in FIG. 4A.
• the previous embodiment suffers from subtracting two noisy quantities.
• This second embodiment begins with the following equation:
• equation 11 The complexity of equation 11 may be reduced by employing
• Equation (11) The function g of equation (11) is not bijective, which means that for every z found under this equation, there is one and only one ratio E s / ⁇ 2 .
• FIG. 5 shows the function g for the ratio E(x 2 )/E( 1 * 1 ), versus the ratio E /sigma 2 .
• the variance may be found under known methods, such as those described in an article by M. Reed and J. Asenstorfer, entitled “A Novel Variance Estimator for Turbocode Decoding,” in Int'l Conf. Telecommunications, Melbourne, Australia, April 1997, pp. 173-178.
• the mean SNR estimate will be less than the true SNR.
• Some bias occurs, due to the fact that deviations exist between the fit of the second-order polynomial estimate and the actual function of the sample averages.
• the curve fit does not apply for very small SNR, nor very large SNR.
• the SNR is very high, then the decoders 304 and 306 can readily decode the frame.
• the SNR is very low, then the frame is lost anyway. If the SNR is in-between, a good estimation of the SNR helps the decoders 304 and 306; so the curve estimates the SNR in the range of about -4 dB to 8 dB.
• the traffic channel need not have a very high signal to noise ratio; however, a good estimation of the energy of the traffic symbols and the associated noise is quite difficult to obtain with only the incoming uncoded bits:
• the energy of the traffic channel over the forward link sent by the base station 106 is changing by plus or minus half a dB after each power control command.
• the TIA/EIA/IS-95-A CDMA standard requires sixteen power control commands during each frame (with 50 frames per second), at 9600 kb/s, where the energy of the traffic channel transmitted during one power control bit group remains constant.
• the IS-95 standard, as well as other modulation schemes employ a pilot channel for transmitting pilot symbols from each base station. The energy of the pilot symbols is constant for a given base station 106. As a consequence, the ratio between the pilot's energy and the traffic's energy is constant during one power control bit group.
• pilot symbols may be found, for example, in co-inventor, Dr. Holtzman's U.S. Patent Application No. 09/144,402, entitled “Method and Apparatus for Reducing Amplitude Variations and interference and Communications Signals, Such as in Wireless Communications Signals Employing Inserted Pilot Symbols," filed August 31, 1998.
• a better approximation of the traffic's energy could be obtained from estimating of the pilot symbols' energy, which is more accurate than estimating the changing energy over the traffic channel.
• the pilot channel can be used to provide a separate estimate for the noise variance ⁇ .
• equation (10) may be simplified by first determining a separate estimate of ⁇ , and then estimating E s from the top half of equation (10) using an estimate of E ⁇ x ⁇ from the signal or traffic channel, or estimate E s from bottom half of equation (10) using an estimate of E ⁇ x 2 ) from the signal or traffic channel.
• a second estimation of E s can be done.
• E p is divided by the approximation of the ratio to find E ⁇ .
• E p is stronger than E s (typically a fifth of the base station power is allocated to E p ), and thus is more accurate.
• the second embodiment described above uses first an approximation of equation (10).
• the received energy of pilot symbols and the ratio between the pilot's and the traffic's channel energy provide a finer estimation.
• a routine 600 performed by the estimator 302 is shown.
• the estimator 302 receives a sample of the incoming traffic signal. Such a sample is greater than or equal to one symbol, and can be a frame or a power control group.
• the estimator 302 fits a stored curve to the traffic channel sample.
• the stored curve may be, for example, the curve of FIG. 5.
• Such a curve may be stored in the memory 218, and the receiver system 204 or control processor 216 may access such a curve for the appropriate curve fitting.
• the estimator 302 determines that the initial estimate of the energy per symbol E s of the traffic channel, and determines the estimate of the noise variance of ⁇ .
• step 608 the estimator 302 estimates the ratio of the energy of the pilot versus the energy of the symbol using the received pilot channel sample.
• the ratio in step 608 may estimated as noted above, where the current ratio is calculated as follows: the current ratio is equal to ⁇ half the previous ratio if the current ratio is higher or lower than the previous ratio, respectively.
• step 610 the estimator 302 determines improved for E s and ⁇ from the ratio E p /E s . Thereafter, the routine 600 loops back to step 602 where another sample is received and processed.
• routine 600 and other functions and methods described above may be performed by the estimator 302 and /or the control processor 216, where the estimator 302 is implemented by a custom ASIC, by a digital signal processing integrated circuit, through conventional logic circuit elements or through software programming of a general purpose computer or microprocessor (e.g., the control processor 218).

Landscapes

• Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Mobile Radio Communication Systems (AREA)
• Noise Elimination (AREA)
• Reduction Or Emphasis Of Bandwidth Of Signals (AREA)
• Transceivers (AREA)
• Error Detection And Correction (AREA)
• Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)

Priority Applications (7)

Applications Claiming Priority (2)

Publications (1)

Family

ID=23165004

Family Applications (1)

Country Status (10)

Cited By (1)

Families Citing this family (19)

Family Cites Families (4)

• 1999
  • 1999-04-29 US US09/301,813 patent/US6393257B1/en not_active Expired - Lifetime
• 2000
  • 2000-05-01 DE DE60035269T patent/DE60035269T2/de not_active Expired - Lifetime
  • 2000-05-01 AT AT00928770T patent/ATE365414T1/de not_active IP Right Cessation
  • 2000-05-01 BR BR0010011-0A patent/BR0010011A/pt not_active IP Right Cessation
  • 2000-05-01 JP JP2000616177A patent/JP2003514408A/ja active Pending
  • 2000-05-01 WO PCT/US2000/012028 patent/WO2000067439A1/en active Search and Examination
  • 2000-05-01 AU AU46952/00A patent/AU4695200A/en not_active Abandoned
  • 2000-05-01 CN CNB008068577A patent/CN1238999C/zh not_active Expired - Fee Related
  • 2000-05-01 EP EP00928770A patent/EP1173959B1/en not_active Expired - Lifetime
  • 2000-05-01 KR KR1020017013857A patent/KR100748286B1/ko not_active Expired - Fee Related

Non-Patent Citations (3)

Also Published As

Similar Documents

Legal Events